In earlier decades, it was assumed that the editor of an edition or staff members directly under her or his direction would keyboard or otherwise transcribe source texts to produce working editorial texts. As in every other area, twenty-first-century methods challenge this assumption.

In the early 1990s, grants from the Packard Humanities Foundation made it possible for editors of the “Founding Fathers” group (the Papers of Franklin, Washington, Jefferson, Adams, and Madison) to mount an effort to complete the transcriptions of documents earmarked for their editions, and to ensure that the machine-readable transcriptions would be available to the projects in searchable form. Some of these projects assigned initial transcription to off-site workers, who received photocopied images of the original documents along with detailed instructions, but who keyboarded the new texts without further direct consultation with the editors themselves. These transcriptions, of course, received careful verification from the editors. Editors were pleasantly surprised at the results, and today several projects use service bureaus for preparation of draft transcriptions. Some of the work is done not only off-site but offshore.

This method demands, of course, even more careful planning of transcription policies and procedures. The Jefferson Papers Retirement Series, for instance, created a basic template, which included rudimentary tags for transcribers working for a vendor whom they employ.

Similarly, OCR (optical character reader) scanning rather than rekeying may be used to prepare a machine-readable transcription of a printed source text. Should the source text present an unusual or archaic typeface, an entirely new transcription may prove more economical. In either case, the transcription must be viewed as a very rough one, demanding serious verification and proofreading by senior staff.

The editors of the Papers of Eleanor Roosevelt have introduced a novel and useful assignment of transcription responsibilities for their edition. Each document is transcribed independently by two different editors. Computer comparison of the two products identifies letters, words, or phrases that demand special attention. Once necessary corrections reconcile the two versions, the transcription is proofread against the source text by still another editor.

Whenever possible, foreign-language materials should be transcribed by someone familiar with the language in which the texts are written. Better still, they should be transcribed by someone familiar with usage of the historical era in which the documents were created. Orthography, spelling, and usage in eighteenth-century Spanish, French, and Dutch are quite different from modern forms of these languages, just as they are for English. While proofreading transcripts against the source texts will catch errors, it’s easier on all concerned to have as reliable a transcription as possible at the beginning.

II. General Rules for Transcription

“Exact copying,” Edith Firth reminded her fellow editors, “involves a fair amount of hack work.” There is no way to make transcribing fun, but there are some simple rules that can make the process less painful and avoid duplication of effort and unnecessary rekeyboarding.

Whenever possible, editors should consult their publishers before transcribing begins. Modern computer typesetting makes the retention of details of inscription like raised letters or archaic symbols much easier, but different programs use different codes for these details. Employing the codes appropriate to the publisher’s system from the beginning reduces the number of corrections and changes to be made later in the editorial process.

When the edition is not a single-source project and has created a database to maintain control of cataloged materials, that same tool can aid in regulating and recording the transcription process. Fields in each document’s record can be used to enter the name of the transcriber and the dates on which transcription, emendation, and proofreading have occurred as the new translation of the source moves through the editorial process.

At the risk of insulting our readers’ intelligence, we offer these commonsense rules:

1. Happily, we can ignore the problems of bygone days created by typewritten carbon copies of transcriptions. With computer equipment, it is still wise to maintain a separate file of proofread but unemended transcriptions and to store backup disks carefully. The editor’s working copy may be a separate file with a different file name or a printout made from that archival file.

Computer disks have not been tested for longevity, and even mainframe computers can lose stored materials. The truly cautious editor may wish to print out two copies of each computer-created transcript, one for editorial use and the second as insurance should the computerized storage facility fail. As editorial work progresses, backup files should be made for each level of revision, and disks with these files should be stored securely and separately. Modern equipment makes the process far easier, and most editors can do automated backup of all their files, while servers can be backed up remotely. With a content management system, all versions are backed up every time a change is made, so that editors don’t even have to think about naming files and ferreting out the location where the most recent copy is saved. The biggest issue for projects may be a system of naming the most recent version of the file and recording where it is stored. Content management systems do this automatically.

This decision is determined by whether editors work on desktop PCs where they store all of their material or whether the project has a central server for storage and backup. When editors store backup materials on their individual computers, there is no secondary backup, and other staff members may not be able to access the most current version of these files.

2. Transcribers should be instructed to insert a special code in the transcript for material that they cannot read in the source. The editor can then do a computer search for this code in every transcription.

3. Transcribers should give their work a preliminary review before filing the copy.

4. The physical format of the printout should reflect its intended use as working copy for an editor. Like any such manuscript, it should be double-spaced, with generous margins on all four sides.

5. Many of a word processor’s convenient features must be turned off for the duration of the project. These include justified right-hand margin. Transcriptions should introduce no new punctuation in the form of end-of-line hyphenation, and ragged right-hand margins on the typescript are a small price for accuracy. More exotic features like AutoCorrect and automatic capitalization must also be disabled.

6. The transcriber and editors should not use a computer font like Times Roman that is proportionally spaced. A nonproportional or “fixed” font like Courier will make it easier to see spacing problems in the transcription.

If transcription can wait until the editorial collection is complete, it’s often more efficient to assign each transcriber a chronological sequence of source texts or a group of topically related sources. The transcriber can thus become expert in the problems peculiar to the time period or theme in question, as well as with the correspondents involved.

III. Special Transcription Methods

The editor who aspires to the CSE emblem must adopt methods to facilitate the preparation of records of editorial emendations required by that agency. Literal transcription of words, of course, is a prerequisite, but the transcriber must also transcribe documents in facsimile fashion in certain areas:

1. The transcriber should preserve the end-of-line breaks of the original document so that the editor can create an accurate record of the author’s end-of-line hyphenation.

2. In transcribing paginated source materials such as notebooks or journals, it is useful to transcribe not only line for line but also page for page. A new page in the document will call for a new transcription sheet so that the editor will have a convenient record of such breaks.

3. The lines of transcription, as well as their pages, should be numbered so that the editor can prepare textual notes without the use of footnote numbers. Most word-processing software offers line numbering as a routine option.

Documents that are entirely in shorthand, codes, or ciphers are usually best left untranscribed until they can be translated. Keyboarding page after page of numerals or alphanumeric codes is an unreasonable and unconscionable trial for a transcriber. Transcribing shorthand symbols would be impossible. Proofreading the results to present to the outside expert or editor who will do that translating would be a ridiculous waste of time. A duplicate hard copy of the shorthand or coded pages will be working copy for the translator. Results of the translation can be entered interlinearly, and the results emended according to the edition’s rules for such passages.

IV. Transcription Forms and Control

In past years, specially designed transcription sheets were almost a necessity for projects that aimed for CEAA/CSE approval. Those sheets carried line numbers along one margin, and their standardized headings included spaces in which to record the completion of special procedures demanded by the CSE: verification, or perfection of the transcript against the original document; completion of requisite proofreadings; and so on. Computer-assisted systems are a special boon here, for such details of format can be entered automatically.

Even projects without expectations of CSE approval learned the value of establishing appropriate formats for their transcriptions, and computer equipment makes the generation of special headings simple. The transcription’s heading should indicate the document’s title, preferably in the form that will appear in the printed edition. The transcription should also bear the project’s code for the location of the original of the source text, as well as the identifying number assigned to that version. Each succeeding page of the printout should repeat the document’s date and identifying source information as well as the new page number. Word-processing equipment makes creation of such headers easy and routine.

Each transcription will require its own unique identifying number. For a project that has collected materials, this can be based on the accession number assigned the photocopy, image file, or recording. When an edition is based on an archival collection, and no identifying numbers have been assigned to the sources at the beginning, the word processor can be programmed to assign sequential numbers to each transcription. Sequentially assigned transcription numbers may also be called for when source texts are transcribed from microfilm reels.

Use of automated equipment for logging transcriptions has many advantages. A system that links data in the control file allows easy and accurate transfer of that information to the transcription record for the same document. Editors can design in advance any forms they or the transcribers will need and can store these formatted headings. When needed, they can be called up by a word-processing macro command. It will be easy to match transcriptions of variant versions of a text with their sources, and if several transcribers are at work simultaneously, each can access the central database to determine which sources need attention and which are already being keyboarded. Later reviews of the transcription can be indicated in fields for the initials of the editors and the duties performed, such as verification, annotation, final proofreading, rekeyboarding, and so on. Database logs of transcription and other editorial processes eliminate the need for hundreds of file drawers and notebooks in editorial projects.

Linking transcription files to the project’s control file system may be the point at which many editors rethink early decisions to use content management systems, relational databases, flat-file databases, or still less sophisticated computer methods. An edition with a few hundred transcriptions may be able to manage without a relational database, but larger ones may have to investigate one. Projects in which a number of editors and transcribers work simultaneously on transcriptions and notes may need networking equipment and a content management system so that each member of the staff can access information when necessary. While access should be open to all, there should be strict control over the entry of information in the record of a document’s progress through publication, to ensure consistency and currency of record keeping.

Nonarchival editorial projects with large staffs and no facilities for networking their computer workstations often find it convenient to file a copy of each transcription’s printout in the file that holds the photocopied source. This provides all concerned with easy, logical access to the transcriptions in progress. Naturally this is impossible when the source texts are original manuscripts that could be damaged by chemical reactions to the transcriptions’ paper or inks or when the source texts are on microfilm. Other projects simply store hard-copy printouts of transcriptions in notebooks, file drawers, or boxes in the chronological or topical order in which they will appear in the projected publication.

V. Factors Affecting Transcription Methods

The same factors that determine the choice of source texts dictate the appropriate method of their transcription. The first are the methods by which the source texts were originally inscribed, printed, or otherwise recorded. The second are the forms of documentary evidence that the sources represent. Only after editors have mastered the intricacies of all the methods used to inscribe the sources in their editorial collections can they begin to rule out textual practices that would distort the details of those sources. The form of documentary evidence represented by each source text—letter, diary, state paper, scientific treatise—will raise another set of questions. The editor should consider the method of communication that each embodies before reaching a final decision on its treatment in the new edition. Whether the editor is his or her own transcriber or delegates this task to a staff member, work will move more smoothly if transcription follows carefully thought out decisions that can be applied consistently rather than choosing a procedure so early that its rules must be changed over and over again.

A. Methods of Recording a Text

The means used to record documentary evidence are often the primary factor that an editor considers in determining standards of transcription and in settling on the final textual method. Certainly it is always the first factor to be weighed. Editors stress the need to master the inscriptional history of unprinted documentary sources, as well as the history of printed materials that will be the basis of a documentary edition. While choosing among competing source texts, the editor should learn as much as possible about the methods by which these variants were produced.

The knowledge may run the gamut from the practices of a group of clerks in the same office to the peculiarities of a particular eighteenth-century letterpress or the details of book production or newspaper composition. For comparable discussions of the inscriptional and organizational history of nonprint sources, see the Eisenhower Papers, “Essay on Primary Sources” (9:2259–73) and “Notes on Primary Sources” (11:1521–23, 13:1507–11).

Unfortunately, there are few secondary studies of modern methods of nontypeset inscription to which an editor can refer. In European medieval studies, there is the recognized academic discipline of “diplomatics,” which offers systematic studies of record-keeping methods of particular groups of clerks and administrators. There are no such formal courses from which to learn how the secretary of the American Continental Congress maintained his records in 1785 or how elementary students were taught to standardize the forms of personal correspondence in 1830. Each editor must learn these methods for him- or herself. Often the best introductions appear in existing scholarly editions, and the novice should consult one that drew on sources similar to the one he or she hopes to publish. Documentary editors who have survived the agonies of learning the intricacies of American inscriptional history frequently share this wisdom in their own edited series. The general rules to be drawn from those experiences follow here.

1. Handwritten Source Texts

Anyone who has transcribed handwritten materials or proofread the resulting copy knows that no typeface can reproduce all the subtle distinctions in the originals. Any typed or printed transcription of such a source is a critical one, because it silently incorporates dozens of editorial judgments and decisions. The editor cannot indicate every instance in which experience enables her or him to recognize a scrawled mark as a period instead of a comma or every occasion when skill and training enable the editor to recognize a slightly inflected line as an n instead of an m. Often the experienced editor exercises such judgment quite unconsciously. Knowing this, the editor should choose a textual policy whose conventions do not conceal additional subtleties in a source that is already being transformed from script to type.

Here we refer you to an exceptionally useful article in the 1999 number of Studies in Bibliography, “A System of Manuscript Transcription,” by David L. Vander Meulen and G. Thomas Tanselle. We thank them, first of all, for a sensible and practical definition of the process of transcribing handwriting: “The effort to report—insofar as typography allows—precisely what the textual inscription of a manuscript consists of” (201). Beyond that, the authors propose an easy-to-use system of recording details of a manuscript that includes neither complicated symbols nor exotic typographic coding. These descriptive codes, of course, can be changed later in the editorial process when other symbols need to be substituted.

Only a careful analysis of the sources at hand enables the editor to decide which conventions will least distort the source. Some writers may employ one method of punctuation in correspondence with close friends and another, more formal, system in letters to strangers and professional colleagues. Obviously, standardizing such marks would conceal important evidence of the author’s relationship with each correspondent. Still other writers vary the length of a standard mark of punctuation such as the hyphen for different functions. When such differences are identified as important, they are marked or coded in the keyboarded transcription to reproduce the author’s consistent, if idiosyncratic, patterns. Individual patterns of capitalization, the use of contractions, and care or carelessness in spelling often provide unexpected insights into an author’s state of mind. The editor cannot responsibly ignore such slips of the pen until certain that they do not represent significant patterns in the writer’s orthography.

For this reason, it’s useful as the project progresses to create a profile for each writer frequently represented in a collection of correspondence or other groups of materials by separate authors. The profile can (and probably should) include images of examples of each author’s idiosyncrasies. There should be just as careful a record of editorial decisions on how to deal with these peculiar styles of punctuation, capitalization, or formation of letters of the alphabet.

It is difficult enough to discover, in a single individual’s handwriting, patterns that change with a writer’s progressive education or with advancing age. The possibility of distinguishing and analyzing such patterns in a group of documents representing a number of authors is far more challenging. Editors who are tempted to impose arbitrary, often modern notions of punctuation or spelling to normalize the two sides of a handwritten correspondence—journals that may have been kept by a dozen scribes, or reports and papers from a hundred contributors—should consider the consequences of such a practice. Such standardization can mean the loss of information about their subjects that readers of the edition might find informative.

Imagination and inventiveness are called for here. One of the thorniest transcription issues faced by editors of the Papers of Margaret Sanger was her patterns of capitalization. Sanger’s script makes every word that begins with th look as though it’s capitalized; she formed initial c’s larger than other letters of the alphabet; and she always capitalized the word “Doctor.” By comparing handwritten materials with Sanger’s use of capitals in typewritten correspondence, it was possible to uncover her true authorial intentions. Thus the edition capitalizes words beginning with c and th when appropriate in standard usage. “Doctor” remains capitalized as Sanger had written it; this practice is evidence of her personal style. When doubt persists, the editors follow standard usage, since Sanger generally did not capitalize words for emphasis, preferring double or triple underlining in such cases.

The process of translating handwritten text to print once almost necessitated standardization. Such alterations to details of inscription as lowering superscript letters were necessary to keep the costs of producing the print edition within bounds. Happily, there’s no longer any compelling reason to adopt such regularizing conventions in the text solely for the sake of economy. Modern computer-assisted technology, including word-processing programs and XML tagging, makes more faithful transcription and publication of handwritten sources in a print edition more affordable. Transcribers can record and reproduce superscripts and subscripts, dashes of various lengths, and other details, producing transcriptions of source texts as literally as their legibility allows. It will be up to an editor, of course, to decide which of these details will appear in the published edition; but computer composition of printed pages makes retention of such details a practical possibility, and traditional emendations like expanding the “tailed p” are no longer an economic necessity.

Printers seem more liberated than ever in their ability to tap into Unicode and import graphic representations for characters for which there is no known symbol, and the users of documentary editions have proven remarkably adaptable in digesting such details. Several well-established projects, such as the Papers of Thomas Jefferson, the Papers of Benjamin Franklin, and the Adams Papers, have reversed earlier decisions on standardization of inscriptional detail with no signs of mutiny from compositors, publishers, or readers. Online editions may need to show more caution: editors of electronic editions must remember that the limitations of HTML display will be significant factors owing to variations in the Web browser being used.

The age and condition of the manuscripts that bear the author’s script may make even a rough transcription of their contents from surrogate images difficult. In such cases, editors should verify transcriptions against the originals before even beginning an assessment of the importance of each detail of inscription. They may have to refer to those originals again and again during the period in which they labor to establish their texts for the edition, sometimes resorting to ultraviolet light to pick up faded ink. If access to the originals is impossible, the tricks of digital imaging available in programs like PhotoShop or Adobe Acrobat can help by enlarging or rotating images or even reversing the direction of text for the deciphering of ink-blot impressions of words.

2. Typewritten Documents

Only in recent years have documentary editors confronted the textual problems created by modern office machines. A recipient’s copy of a signed typewritten letter or other communication presents few problems: no matter what its flaws, it may be considered to represent the author’s final intentions. If the typescript carries the author’s handwritten corrections or revisions, these can be indicated by textual symbols, special typefaces, or annotation. Often the same methods that the edition adopts to meet the needs of handwritten sources can be modified for typed sources. In the absence of such symbols, the editor instructs the transcriber to key in such emendations in a special typeface reserved for handwritten additions to typed pages or to flag them with a special code. If such authorial revisions are rare, footnotes can explain these occasional details of inscription. Whatever the final decisions may be, the initial transcription should indicate all these details, so that they can be recorded as the formal textual apparatus dictates.

An uncorrected carbon copy creates new problems. Faced with file drawers full of such uncorrected copies of outgoing correspondence, the editors of the Woodrow Wilson Papers compared surviving carbon and ribbon copies of the same letters and found that Wilson himself had corrected the originals. They concluded that the ribbon copies of these letters had been corrected before dispatch, and the texts were emended accordingly. The Eisenhower edition took a similar approach. After analyzing Eisenhower’s patterns in proofreading outgoing typed correspondence, the editors adopted a twofold policy of emending errors in typed letters for which only the retained carbon copy survived. Mere typographical errors and errors of spelling, which Eisenhower routinely corrected, were emended silently, but editorial corrections of misspellings of proper nouns are accompanied by a footnote recording this emendation.

Editors of late nineteenth-century typescripts should exercise special caution. Any assumptions about “typographical” errors here must take into consideration the specific keyboard of the machine employed. Decades passed before these keyboards were standardized, and emendations can be made only with full knowledge of the particular instrument involved. Contemporary photographs of the subject’s office, bills, or correspondence ordering office equipment help the editor to identify the machine employed. If the machine’s manufacturer is no longer in business under the same name, the editor may have to embark on a new career as corporate historian to find a source for an image of the typewriter’s keyboard and determine patterns of likely typographic error. If no such records survive, the editor may have to abandon plans to correct such errors wholesale, for the identification of their patterns may be well nigh impossible.

Transcription and textual policies for typed sources should reflect not only that medium of inscription but also the basic nature of each source. A typed draft letter with handwritten or typed revisions has requirements quite different from a nonauthorial typed transcription of a handwritten source. Editors who ignore this rule and lump together typed letters, retained carbon copies, typed transcriptions, and printed transcriptions of lost sources into one textual category compromise their product before the process of transcription even begins. Typewriting is merely an exceptionally legible form of inscription by an individual, and the appearance of a typed document should not mislead an editor into ignoring other factors that may dictate editorial treatment.

3. Printed Source Texts

The editor who establishes the text of a published work for the purposes of a critical edition may sometimes legitimately depart from the substantives and accidentals of any single surviving printed copy of that writing. But the documentary editor views print as merely another way to inscribe a document, and the surviving copies of a given source must be evaluated to determine which is the best document, the one that is a unique, authoritative source with evidentiary value. Once that source has been identified, it should be transcribed literally and published without emendation, conflation, or the other heavy artillery of textual editing.

4. Nonverbal Documents or Document Elements

The editing of completely nonverbal documents was not addressed until late in the twentieth century by editors in the NHPRC and CSE traditions. Earlier, many editors had to solve the problem of how to present scattered authorial illustrations and even doodles that were part of a more conventionally inscribed source. When these elements had recognized equivalents in type, the transcriber entered such ready-made symbols from the printer’s font in good conscience. The editors of the Hawthorne Notebooks translated the author’s hand-drawn marginal “fists” into ☞, the existing standardized type unit for this symbol (called a “fist graphic”). The editors of the Emerson Journals employed the conventional typeset caret (ˆ) for Emerson’s mark for interlineation; and Mark Twain’s editors translated Clemens’s autograph proofreading symbols in his Notebooks and Letters into the equivalent printer’s symbols.

Unfortunately, writers seldom confine themselves to such easily translated marks. Some authors, like William Blake, were artists as well, and neither critical nor documentary editions of their works would be complete without reproduction of the images that are an integral part of their texts (G. E. Bentley, “Blake’s Works as Performances: Intentions and Inattentions”). With modern print publication techniques, the photoreproduction of sketches, drawings, doodles, or maps in a printed volume of documents is far less expensive than it once was, but the method is still limited to occasions when the literal presentation of such elements seems essential to understanding the documents. (See, for instance, the facsimile of a Jefferson diagram in Papers of Thomas Jefferson, 22:74.)

Among documentary editors, scholars in the field of the history of science and technology seem more ready to explore the transcription of nonverbal materials—and to report results to their peers. In the September 1984 issue of Documentary Editing, Reese Jenkins and Thomas Jeffrey of the Edison Papers pointed out that most editors had actually retreated from these challenges. Julian Boyd, they noted, relegated Jefferson’s nonverbal documents such as architectural sketches to a separate series—one that has not even been started although the Jefferson project is well into the second half of its first century. The Franklin Papers excluded nonverbal materials altogether except as illustrations. And although the Emerson editors reproduced Emerson’s easily translatable symbol for an insertion, they excluded all his pen and pencil sketches from the Journals and Miscellaneous Notebooks (Jenkins and Jeffrey, “Worth a Thousand Words: Nonverbal Documents in Editing,” 7).

In happy contrast, a more recent examination by the editors of Charles Darwin’s Correspondence shows the increasing ease with which they can deal with the phenomenon of “extra-text” objects such as tables, formulas, photos, and diagrams. In 2004 they reflected on the advances in print technology that allow them to abandon photographs of photographs or even tracings of photos in favor of high-quality digital scans of original illustrations for their editions (Duncan Porter and Alison Peran, “Editing Textual and Extra-Textual Materials in Charles Darwin’s Correspondence”).

Still, as Jenkins and Jeffrey first noted more than twenty years ago, editors of papers in the arts and sciences have a better record for the treatment of nonverbal documents than editors of the papers of a statesman or a novelist. The graphic records of great artists and illustrators introduce a new dimension to the problem of editorial transcription. Reproduction of such records for author-artists must meet technical standards far higher than those adequate for the comparatively simple line drawings of a scientist or the doodles of an author in a notebook. American documentary editors first addressed the problems of reproduction techniques in editions of the papers of figures like Benjamin Henry Latrobe and Charles Willson Peale. In these two examples, the costs of reproducing all sketches as a physical part of the printed pages would have been prohibitively expensive. Instead, separate microfiche supplements were published to provide students with a comprehensive reproduction of graphic materials, while the book editions reproduced selected drawings and sketches on plates inserted as illustrations to the printed text.

Web-based publication makes unnecessary many such agonizing decisions about inclusion and exclusion. The online archives of William Blake (http://www.blakearchive.org/blake/) and Dante Gabriel Rossetti (http://www.rossettiarchive.org/) provide easy worldwide access to an astounding variety of images. For most editors, image editions like these will be unnecessary. However, almost all will deal with an instance when the nature of a sketch or drawing or marginal doodle is essential to the reader’s understanding of a document’s verbal elements. Even when literal photoreproduction of the sketch is not required, some editorial action is necessary. Transcribers will often be the first to notice such symbols, and there should be a code that can be entered to bring such images to the attention of an editor. If they are comparatively rare and isolated, the editor may choose to describe them in a note adjacent to the text or simply import the graphic into the text. A recent example of this is the marginal reproduction of Thomas Jefferson’s sketch of a chair in Papers of Thomas Jefferson: The Retirement Series 1:44.

If such graphic materials are common to the group of documents being edited, the editor must find another course. Some may need to imitate the Latrobe and Peale editions by issuing facsimile image supplements where the reader can consult the unprinted and even unprintable elements of the documents. The diagrams and schematic drawings of great figures of science obviously fall into this category. Joseph Henry’s sketches and diagrams were conscientiously reproduced in the Henry Papers, as are such elements in laboratory notebooks and other technical sources in the Papers of Thomas A. Edison and Albert Einstein. Here the “transcription” of the source text is a photoreproduction, as much a translation of the original as a typescript is of a manuscript letter. The nature of the original (an artist’s oil painting or an inventor’s line drawing) and the needs of the audience (art historians or students of the history of technology) dictate whether that translation needs to reproduce the original’s colors or subtle shadings or whether a low-resolution black-and-white scan will suffice. When the “transcription” is a photograph of a three-dimensional artifact, the gap between source and editorial text is even more obvious, and edited nonverbal texts can present problems analogous to verbal sources at almost every stage. For instance, an editor may have to emend a photographic transcription by retouching it for photoreproduction in a print or facsimile edition or create a critical “clear text” of the image by creating a new schematic redrawing of an original technical sketch so crude that it failed to convey its creator’s obvious intent.

Much of the most useful literature on this subject comes from the editors of the Edison edition—Reese Jenkins’s “Words, Images, Artifacts and Sound: Documents for the History of Technology,” and Robert Rosenberg’s “Technological Artifacts as Historical Documents”—but the discussion has now enlarged significantly. Jenkins and Rosenberg examined the relationship between artifactual documents and recorded sound. Jenkins conceded that this might seem a special concern for historians of technology: “To ‘read’ the nonverbal artifactual record is to read what the inventive enterprise was all about. In the case of the Edison papers, to have a documentary edition without the artifacts would be equivalent to having a marriage without the bride” (45). But Jenkins and his colleague Jeffrey argued persuasively that all editors must now confront the challenges of nonverbal documents: “While acknowledging that our Western historical tradition was rooted from the beginning in words, we must also realize that the documents of the past, and, increasingly, of the present and future beckon for recognition and challenge us with diverse languages expressed in symbols other than words” (Jenkins and Jeffrey, “Worth a Thousand Words: Nonverbal Documents in Editing,” 8).

5. Maps

While most documentary editors use maps only as helpful illustrative adjuncts to their texts, some have had to confront the question of editing maps as historical documents. Stephen E. Wiberley Jr. offers a detailed discussion on the cartographic equivalents of facsimile publication, emended transcriptions, and clear text in “Editing Maps: A Method for Historical Cartography.” Wiberley’s discussion centers on his experience with the Atlas of Early American History. A more strictly documentary approach to maps as evidentiary sources was pursued in the Atlas volume of The Journals of the Lewis and Clark Expedition.

B. Transcribing Different Types of Records

The method of transcription should reflect not only the physical format or history of the document but also the type of documentary record that each source represents. The traditional literary distinction between a writer’s public and private writings is of little use to the documentary editor. Unpublished letters have influenced the course of history. State papers not set in type in their author’s lifetime have shaped the thinking of legislators and executives. Confidential technical reports have radically affected the history of science. Each of these is private under the literary scholar’s definition, but all were composed for an audience, and all derive their historical importance from the influence they exerted on those contemporary readers.

Thus documentary editors long ago abandoned any methodological division between public and private writings and, instead, examine how these source texts functioned as agents of communication. Any edition of such sources should strive to preserve the communicative intention or effect of the original. Recent scholarship in the humanities as a whole, of course, seconds this decision.

With documents, the editor may also have to modify notions of what constitutes the appropriate contents of an edited text. Elements in the source that could safely be ignored by the editor of an author’s published works may be an integral part of a source’s documentary contents. Such accidentals as capitalization, indentation, and spacing may perform important functions in the design of certain modes of communication. And for some documents, even nonauthorial contributions must be considered part of the source’s evidentiary contents.

1. Correspondence

The literature on editing correspondence once focused on the need to make the letters of an earlier age “readable”—even “enjoyable”—for a modern audience (see, for instance, Robert Halsband, “Editing the Letters of Letter-Writers”; and William Ferraro’s review of Mark De Wolfe Howe’s Home Letters of General Sherman). Today, discussions of the textual needs of correspondence are considerably more sophisticated, even if agreement on methods remains distant (see Robert Stephen Becker’s report on his survey of editions of nineteenth- and twentieth-century letters in “Challenges in Editing Modern Literary Correspondence: Transcription”).

The move toward clear text in CSE-approved editions of correspondence was justified, in part, by the claim that the literary critics who formed a large part of these volumes’ audience would not wish to be distracted by textual symbols or numbered footnotes pointing out details of the original. Even exponents of silently emended, expanded transcription concede that readers soon become used to archaic and idiosyncratic usage in a documentary edition, and Ernest W. Sullivan has pointed out that literary critics were ill served by the practice of standardizing texts in editions of personal correspondence (“The Problem of Text in Familiar Letters”). Steven Meats lectured the editors of the correspondence of literary figures: “As a general rule, the less editing (that is, the less editorial emending and altering) done to the text of letters, the better the job of editing. A letter is, after all, a primary historical document; one might even call it a ‘fact.’ In any case, silent emendation in the editing of letters should be severely restricted” (“The Editing of Harold Frederic’s Correspondence,” 38). As the editor’s compulsion to conventionalize, normalize, or otherwise emend the text of letters came under serious attack, editors of the private writings of American literary figures moved further and further away from the aim of a clear text.

It is not the business of documentary editors to introduce new readings into a documentary text for the sake of historically unrealized clarity. When letters survive in some preliminary form that reflects revision by the author, they may eventually require the application of textual methods appropriate to genetic elements (see pp. 188–91, below) that reflect the evolution of a text, but the initial transcription should be literal. When a letter survives in the form in which its author dispatched it to its addressee, it deserves the same treatment in transcription. Literal transcription of hastily scrawled and ill-proofread recipients’ copies of letters may be untrue to the author’s final intentions, but such transcription is true to the documents. Equally important in documentary terms, it represents the form of the letter that influenced its recipient. Unnecessary emendations will make the editorial text of any letter useless as evidence either of what its author wrote or of what its addressee read. Even editors who are their own transcribers should make initial transcriptions of such sources as literal as possible. Any policy of emendation for letters adopted later should be as conservative as their method of inscription allows. A detailed back-of-book textual record grants no license to violate this rule.

If any corrections are to be imposed on the author’s prose, they will come after the editor has established the edition’s final textual policies. And all editors should remember that most people do not write nonsense. Words that are incomprehensible to the transcriber may in fact be archaic words, creatively spelled words, or foreign terms. “Correcting” apparent mistakes will take the editor down the wrong path. A correction of such words should only take place after all the above possibilities have been eliminated. Editors should consult the Oxford English Dictionary (so easy in the online edition), and foreign-language dictionaries, and consider alternative spellings or similar pronunciations. (For a compelling discussion of the need to remember the effect of punctuation on oral patterns, see Kathryn Sutherland’s review of Chapman’s editions of Jane Austen’s Mansfield Park.)

Transcribers are generally authorized to incorporate certain conventions for standardizing formal elements of letters. Such standardization most commonly concerns the location of datelines, greetings, and closings. Even this degree of intervention may prove shortsighted if the letter writer employs individual formats for different types of correspondence. David Knight, editor of Sir Humphry Davy’s correspondence, pointed out that there was much to be learned from the openings and closings of these letters: “We can trace, for example, through the tops and tails of his letters, his relationship with Faraday as he progressed from amanuensis/valet to what in modern terms might be ‘research student,’ to ‘research assistant,’ and on to colleague” (“Background and Foreground: Getting Things in Context,” 10). Should these variations show a significant pattern, the original format should be maintained.

When the writer used stationery with an imprinted letterhead, there is no reason to repeat this form at the beginning of the transcription of every letter inscribed on such paper. The editor may reproduce the letterhead’s text verbatim or simply import its image as an illustration at some point in the edition, and design an easily recognized contracted form that can be used in the printed edition.

a. Source or Provenance Note Information

The letter’s transcriber is also responsible for recording information that may eventually find its way into the document’s source note or provenance note (see pp. 227–31, below) rather than into the editorial text. If the letter’s envelope or address leaf survives, all authorial inscriptions should be transcribed verbatim. At the very least, the transcriber should note postal markings. When they represent the basis for assigning a date to an undated letter, they should be described in detail. Any notations by postal officials or others involved in forwarding a letter to its intended recipient should also be noted. Once these elements have been transcribed, the editor will review them and decide on their treatment in the edition.

The transcriber should also indicate the existence of any endorsement, that is, a notation made on the letter, its envelope, or its address leaf after its receipt. Whenever such an endorsement indicates the date of the letter’s arrival, summarizes the reader’s reaction, or otherwise supplies important documentary evidence, it should be transcribed verbatim. Correspondence with governments or their agents often carry endorsements made after the letter’s receipt by someone other than its addressee and other important documentary notations. They, too, must be recorded in full.

These suggested practices apply only to sources that are truly letters, not essays or other short works written in letter form with the intention of print publication. The problem of distinguishing between the two is discussed in John A. Walker’s “Editing Zola’s Correspondence: When Is a Letter Not a Letter?”; John M. Robson’s “Practice, Not Theory: Editing J. S. Mill’s Newspaper Writings”; and E. Grace Sherrill’s “ ‘The Daily Crucifixion of the Post’: Editing and Theorizing the Lowry Letters.”

2. Business and Financial Records

Documentary editors generally agree that documents such as business records, accounts, and others recorded in a tabular manner should be printed in the most literally transcribed format possible. The reasons are twofold. First, such documents make sense in visual terms only if the original arrangement of columns and indentations is preserved. Within such a format, it is usually impossible to expand the author’s contractions or abbreviations to make headings or entries fit into the spaces available. Second, there is no theoretical justification for emending such records in the name of recovering the author’s literary intentions. Such sources have only evidentiary, documentary value, and there is no reason to emend them to achieve easily appreciated literary value.

3. Professional and Technical Records

Editors who publish professional and technical records should not only master the technical terms in the document’s verbal text but also become familiar with the special formats peculiar to such professions as law, medicine, physics, or mathematics. They can assume that a large proportion of their readers will themselves be specialists in the history of these fields. Such readers will be best served by a faithfully printed facsimile of the source’s format, preserving the styled brackets of the attorney and the indentations and spacing of scientific formulas. These elements need not be standardized in the name of readability, for they will already be familiar to the edition’s audience, and their normalization would destroy important aspects of the source as it was communicated to its original audience of judges, court clerks, or fellow scholars.

Yet with scientists and businessmen, just as with statesmen and authors, the line between public and private writing will blur again and again to make the editor’s life more difficult. M. J. S. Hodge’s instructive review essay discussing Charles Darwin’s Notebooks and Correspondence comments on this phenomenon: “It would be a mistake to see Darwin as addressing himself in the notebooks, addressing single, known correspondents in his letters, and only going openly to an indefinitely broader public in the books. For, even when Darwin wrote in his notebooks or his letters, he was articulating thoughts that were formulated and refined in his mind as potential contributions to the intrinsically public activity that natural history and science were explicitly committed to being” (“Darwin: The Voyage, London, and Down”).

4. Government Records

Official records of any government or its agents cannot bear emendations in the name of clarity, readability, or enjoyability. They are what they are what they are. Government financial records, of course, are subject to the rules for tabular documents, but special respect is also due the formats and accidentals of nonfiscal records. Legislative journals, like financial records, are often intelligible only when their original patterns of spacing and indentation are retained. Greater respect is also usually shown the original formats of state papers outside the definition of journals, for these visual and inscriptional patterns were familiar to the lawyer-legislators and lawyer-clerks who inscribed them.

Transcribers should also copy the nonauthorial labels and notations on government records faithfully. Eventually they will be noted, described, or printed verbatim in the final edition. Because such documents are remarkably long-lived in terms of their documentary significance, an unprinted legislative report or treasurer’s account from the 1790s may well have been consulted by congressional committees two decades later. This, in turn, may be indicated by the notations of the clerk responsible for filing and refiling the manuscript. Unless such dockets are transcribed or noted fully in the edition (either as part of the text or in a note), the source’s documentary elements have been only partially reproduced.

5. An Author’s Works

An author’s works—essays, stories, plays, books, or other forms written with the intention of print publication—used to receive different treatments from the textual editor and the documentary editor. Textual editors who hoped to recover an author’s intended meaning sometimes emended one source text or conflated authorial portions of two or more sources into one new text. Textual scholars now look with more favor on the “archival” approach, that of documentary editors, who focus on a single source for their new edition, with the new noncritical editorial text reflecting the characteristics of that one source. If the source is a draft version, it should be transcribed according to the edition’s general rules for sources with genetic elements. If it is a fair copy or a printed version, the transcription should be a verbatim rendering of that single source, with later emendations indicated clearly by the editor and with variants recorded later in adjacent notes rather than incorporated into the new editorial text.

6. Journals and Diaries

The nature of a source’s original intended audience may influence transcription policies and textual method for literary works, public papers, and even letters, but this is not the case for an author’s journals or diaries. Here the intended audience is usually the author (although one wag remarked that in the case of the Adams family, the diarists’ audience was posterity). These intimate records, revealing so much of the inner life of a public figure, demand the most literal textual treatment their method of inscription permits.

Informal in nature and private in intent, diaries lose rather than gain by any attempt to impose excessive conventions of print publication. If a writer’s punctuation or spelling is less regular and correct in diaries than in correspondence, so be it. The very fact that authors allow themselves such lapses may be significant. If the author employs any form of shorthand, though, more liberal editorial intervention is not only permissible but also necessary.

The format of a diary or journal may also require special treatment. Diary entries inscribed in books whose pages carry no preprinted dates will demand additional editorial intervention in standardizing or expanding the dates of entry furnished by the author. The editor’s annotational format may play a part in this decision. When numbered informational notes follow each daily entry, an arbitrarily formalized heading, free of brackets or other typographical barbed wire, may be less distracting; with footnotes, the daily headings are less striking visually. Some standardization is often necessary even during transcription, and the most common method consists of placing each date flush to the left-hand margin. The editor will decide later whether to set the date in boldface or italic type or in full and small capital letters to give the reader easy reference. With such a run-on text of diary entries, it may be necessary to normalize the substantive elements of the date as well as its form. Expanding abbreviated forms in the dateline or even standardizing the author’s arrangement of day, month, and year may be necessary to give the reader consistently useful reference points.

Bound diaries and journals often require a more complete description of the original source than letters and other shorter manuscript sources. If the author has paginated the journal, or if the journal carries preprinted numbers on its pages, these should be noted by the transcriber for reproduction, in brackets, in the new print edition. If the diary’s pages are unnumbered, the editor may need to assign numbers for easy reference (see the explanation of such a plan in the introduction to the Irving Journals, vol. 1).

Editions in the CSE mode tend to carry more detailed descriptions of the physical appearance of bound journals and diaries, but even historical editors recognize the need to offer more descriptive information for this category of source text. As examples, see introductory notes for the John Adams, John Quincy Adams, and Charles Francis Adams Diaries. The reason for such explicit description is practical. The reader must be able to locate the original entry for comparison with the editorial text, and this is clearly a greater challenge within a bound volume of diary entries than in a separately cataloged one- or two-page letter. Thus, the transcriptions themselves should indicate breaks between volumes, as well as note any special problems in arrangement of entries in the original.

7. Records of Oral Communications

Editors transcribe records of the spoken word with special care, for any printed or even handwritten versions will be far removed from their originals. Multimedia electronic editions offering access to both the audio record of spoken words and their machine-readable transcriptions can address this problem fairly simply. (They find, of course, that users who can compare recorded words with their editorial transcriptions are even more impatient with inaccuracy.) Most editions do not have this luxury.

Transcriptions of oral communications present a visual record of words and thoughts intended to be spoken aloud. Only with the greatest skill can such a text aspire to be even second best in documentary terms. The pauses, tonal inflections, and accompanying physical gestures that once gave these words their authors’ full intended meaning are lost, and editors are often faced with conflicting inscribed versions of the same spoken words. Here, documentary editors resort, again, to the critical methods of textual editing to reconcile transcriptions of variant versions (see pp. 187–203, below).

a. Inscribed Records of Spoken Words

Editors of documents created before the development of phonography are fortunate when they find even an author’s draft, outline, or notes for a speech. In the absence of any other record, it is impossible to determine whether such a “pre-text” bears any relationship to the lecture or oration actually delivered.

Whenever a pre-text is the sole source, it is transcribed and edited as conservatively as its method of inscription allows. The question of audience here is a delicate one, for the pre-text itself was intended for its author in visual form, an aide-mémoire to the words and phrases that he or she intended to utter. When a pre-text was scribbled hastily or inscribed in some authorial shorthand, editors should consider expanding idiosyncratic contractions and translating shorthand symbols. The editor should be exceptionally restrained in emending or standardizing such formal elements as spacing, punctuation, and the like, for each device may have served the author as stage directions in speaking, and no useful purpose is served by destroying the clues that format provides. The modern edition of Jonathan Edwards’s History of the Works of Redemption, for instance, retains all of Edwards’s unique marks of punctuation to indicate pauses for emphasis in these sermons.

When the editor has access to records of sermons or lectures made by members of the speaker’s audience, it is usually wise to prepare literal transcriptions of any and all versions that survive—anything from a brief newspaper mention to a full stenographic text. Editorial notes can explain the nature of the source, and readers can judge for themselves how accurately the reporter has mentioned, narrated, or summarized the speech or conversation involved.

Fragmentary paraphrases can also occur in other documents, like a newspaper article on a matter irrelevant to the edition. In such cases, there is no need to transcribe the larger document. The editor customarily omits those portions that do not concern the oral text in question, always indicating omissions with ellipses or some other device reserved for that editorial intervention. (See, for instance, the treatment of “third-party” documents discussed in the introduction to vol. 1 of the Lafayette edition.)

Through the early nineteenth century, various mentions, narrations, and summaries of the same speech or conversation often survived in textually irreconcilable versions. In the absence of systematic shorthand, no two reporters left accounts of the same spoken words that could be viewed as variants of the original. They are usually so dissimilar that there is no question of conflating or combining them into one master record of the spoken words. Instead, each should be transcribed literally, for any normalization of format or punctuation could obscure the meaning of the original that the reporter tried to convey. The editor may be unable to identify one source text, choosing instead to give readers access to all conflicting reports. Should one record of the speech be a ten-page narration while the others are paragraph-long summaries of the same words, editors usually transcribe and print the longest record as the editorial text, reporting transcriptions of the shorter versions in notes or as separate documents.

b. Sound-Recorded Documents

Scholarly editors have just begun to address the textual problems of the newest documentary records, those created by the perfection of sound-recording equipment. When tapes or phonograph discs for actual speech or conversation survive, editors have at hand something close to the archetype for the document—their transcriptions will be the first imperfect witnesses in a long series. The skill or awkwardness with which documentary editors meet this challenge will take the measure of their scholarly specialty. At the moment, the editions paying closest attention to these problems are the Martin Luther King project and the Presidential Recordings series. For King, the editors have as source texts audio recordings of the sermons he delivered at various African American churches. The “text” of these recorded sermons includes not only the words King uttered but also the words of the congregation’s responses. The methods used to translate these elements for print publication are described by Peter Holloran in “Rediscovering Lost Values: Transcribing an African-American Sermon.”

The problems of transcribing oral history interviews have been addressed at length by experts in this field. In addition to books like the Handbook of Oral History and The Oral History Manual, Baylor University’s Institute for Oral History provides a free transcribing style guide on their Web site at http://www.baylor.edu/Oral_History/index.php?id=23607. The editor of oral history interviews should learn as much as possible about the peculiar inscriptional history of this form of document, for the practices of American oral history archives can camouflage pertinent facts. Customarily, both the interviewer and the subject review the typed transcriptions of such memoirs, the first correcting errors of transcription and the second making emendations for style and indicating passages that he or she wishes omitted from the final archival version of the transcript. Some oral history projects even destroy the original tapes, thus eliminating any chance for comparing the typed witness against its archetype. Most oral history projects now use word-processing equipment to emend transcriptions after review. This use of computer technology may ensure that only the final version of such interviews survives in the word processor’s storage, with earlier and fuller versions lost forever. As American documentary editors catch up with the products of phonographic technology, this form of source text will become a focus of lively debate.

C. Transcribing an Earlier Transcription

Sometimes earlier handwritten, typed, or printed versions of a lost archetype are all that survive. Whatever their format, such earlier transcriptions are to be regarded as scribal copies—or copies of copies—of those lost sources. They should be transcribed literally for the modern edition, and any later emendations should be as sparing and carefully considered as possible. Frequently, the editor will be able to second-guess the earlier transcriber, recognizing that the scribe has consistently read the author’s a for o or r for n. Such systematic mistranscription by a specific copyist can and should be emended, but only when the editor’s notes warn readers by explaining the reasons for these decisions and pinpointing the areas where emendation has occurred. Similarly, typographical errors in printed transcriptions (whether based on manuscript, typed transcriptions, or the lost originals) can be corrected sparingly if such emendation is needed to make the transcriptions intelligible. Here, too, later notes should explain editorial decisions in the case of any substantive emendations.

The transcriptions’ punctuation and other accidentals should usually stand without any attempt to make them conform to conventions established for original source texts in the new edition. Second-guessing an earlier transcriber does not extend to mind reading.

VI. Conclusion

While some rules of editing change with electronic technology, some do not. In “Some Unrevolutionary Aspects of Computer Editing,” Hoyt Duggan writes: “Simple accuracy is perhaps the hardest thing to achieve in an edition, and electronic technology scarcely affects the labor of transcribing and proofreading. Transcription on a keyboard, like writing on animal skins with a quill, still takes place character by character.”

Even the most basic rules for transcription of source texts raise the questions central to scholarly editing—the considerations that make possible the establishment of printed texts that reflect the editor’s experience and knowledge. A series of literal, verbatim transcriptions is almost always the sturdiest base for any documentary edition, but the edition’s textual standards as well as its organizational format may require a modification of this rule. The conventions by which American editors have presented documentary texts are analyzed in the next chapter. Well before they begin transcribing source texts, editors need to familiarize themselves with these modern textual conventions, lest their working transcriptions fail to serve the editor’s purposes.

Transcribing the Source Text

I. Who Does the Transcribing?